This Week on perl5-porters (11-17 November 2002)

This Week on perl5-porters (11-17 November 2002)

I finally found time to produce a new instance of this weekly selection. Read about Unicode chars in POD, an incompatible change in a widely (and wisely) used module, a couple of new proposals, and a smaller than usual amount of bugs.


A thread was spawned from the B::Lint patch proposed by Ian Phillipps last week. It's about non-ASCII chars in PODs (for example, ï). Apparently, some versions of nroff(1) are offended by characters with an ASCII code > 127, and pod2man should generate portable manpages (although Russ Allbery said that he was considering a way to add an option to pod2man to generate groff-specific manpages). Sean M. Burke suggested to use the Pod::Escapes module to get rid of escape sequences that can't be handled in the output format, and added some insight about what the POD specification says on Unicode input.

Test::* modules change

Michael G. Schwern released a new alpha version of Test::Simple/More/Builder. An important change is that is no longer automatically loaded on an ithread-enabled perl. Hence it will be necessary to use thread before loading Test::More or Test::Builder to run tests that use threads.

Assertions in Perl

Salvador Fandino send an impressive patch to add assertion support to Perl. Briefly, a new sub attribute assertion is used to mark subs as assertions, and those assertions are enabled via a new command-line switch, -A. Yitzchak Scott-Thoennes found that it's a great idea. The pumpking hasn't commented on this yet.


Brian Ingerson posted an announcement of his CPAN::MakeMaker module on P5P last month, and didn't receive any comments until this week, probably because the implications of this awesome hack are difficult to figure out at first sight.

CPAN::MakeMaker is a replacement for ExtUtils::MakeMaker. It's also self-distributing : it's part of the MakeMaker-generated tarball, hence doesn't need to be installed on the users' systems -- in fact it gets never installed on the users' systems.

Version bug

Andreas Koenig apparently found a bug in Basically he states that

    use version;
    $v = version->new(5.00563);
    print $v->numify;

should print "5.00563", and not "5.005063", since it should be equivalent to 5.5.630, and not to 5.5.63. The correct interpretation of the relevant chapter and verse of perl56delta are required to define the proper behavior.

In brief

Sean M. Burke announced a shinier, refactored, and generally refunkified version of the perldoc utility, available from CPAN if you want to test it. It's of course aimed at replacing the current perldoc.

Bug #18306 shows that B::Xref sometimes reports random file names with a threaded perl. Apparently the problem comes from the perl core, not from B::Xref.

Dave Mitchell sent a patch to add more information to the infamous error message Scalars leaked, by printing the adresses of the leaked scalars. It's enabled by the new preprocessor symbol DEBUG_LEAKING_SCALARS.

H.Merijn Brand still chases Cygwin bugs through a deep smoke screen.

About this summary

This summary brought to you by Rafael Garcia-Suarez. If you prefer a text/plain interface, it's also available via a mailing list, which subscription address is Comments and corrections are, as always, welcome.